669715 - TI: fix ARM bustage

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Brian Hackett [Laid off!]]

The TI -> TM merge broke most of the Android and Maemo mochitests and talos tests (which were not being run on the JM tinderbox, oops).  Disabling TI on ARM fixes this, but this is just a stopgap and we need to figure out what is crashing ARM.

Comment 1

[Brian Hackett [Laid off!]]

Stopgap changeset to disable inference on ARM.

http://hg.mozilla.org/tracemonkey/rev/8570f783088c

Comment 2

[Joel Maher ( :jmaher ) (UTC -8)]

using a build before the disable, I see fennec crashing and this is the crashdump that I see:
http://people.mozilla.org/~jmaher/android_dumps/52b96947-9f60-f7d4-30977068-797f8a56.dmp
http://people.mozilla.org/~jmaher/android_dumps/52b96947-9f60-f7d4-30977068-797f8a56.extra

fyi, this should be in crash-stats, but I couldn't find it.

Comment 3

[Joel Maher ( :jmaher ) (UTC -8)]

I did get this in crashstats:
https://crash-stats.mozilla.com/report/index/5a3875bf-0801-4dfa-8a04-21bbb2110707

Comment 4

[Boris Zbarsky [:bzbarsky]]

This is causing a slowdown for ARM over the pre-TI state of affairs, so requesting tracking for mozilla9.

Comment 5

[Brian Hackett [Laid off!]]

Attachment: WIP

Fix some ARM ABI-related test failures.  Blocked on an ilooping jit-test for now, due to breakage in the ARM assembler.  Filed bug 684404.

Comment 6

[Marty Rosenberg [:mjrosenb]]

Attachment: fix silliness with pools

when flushing a pool, we were not resetting the number of elements since the last pool when there was nothing in the pool.

Comment 7

[Brian Hackett [Laid off!]]

Still more to do here, but now that jit-tests are passing and jstests are almost passing it's good to get some Fennec tinderbox coverage for this stuff.

http://hg.mozilla.org/projects/jaegermonkey/rev/9459096a263a
http://hg.mozilla.org/projects/jaegermonkey/rev/b31b25125429

Comment 8

[Brian Hackett [Laid off!]]

https://hg.mozilla.org/mozilla-central/rev/c943bbf9dac4